Bridging bytes and behavior with Data Science and UX Research
summary
Claims that 120 zettabytes of data are produced last year and that 80% is unstructured while only 2% is retained set stakes. The session maps an actual Data Science pipeline—wrangling, NLP, clustering, anomaly detection, A/B testing—and shows concrete interventions: Spotify anomaly ad campaign, a diary study revealing ad‑skip mental models, ABN AMRO’s automated tagging that initially increases agent time until tags are pruned, Google deep learning at 90% agreement with humans, and Figma’s joint data + UX fixes.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 3:16Speaker and UX overview
- 6:15User research vs data science roles
- 9:08Data science pipeline
- 12:18Machine learning and overfitting
- 17:56Model evaluation and storytelling
- 22:57Quantitative vs qualitative research
- 25:39Spotify anomaly detection case
- 27:12Spotify ad skipping limits study
- 29:18ABN AMRO tagging failure
- 31:25Google AI usability study
- 32:09Figma notification issues
- 36:04Ethics and bias
- 39:39Conclusion
related talks
42mWhy Data Science and UX Research should be Best Friends
19mData science demystified
28mCrafting Hyper-Relevant User Experiences
34mPast The Hype: A Better Way To Think About Big Data
20mDesigning for Big Data
26mData For The 99%: Unlocking Data Curiosity Across Your Organisation
28mThe Value of UX Applied to Company Data
41mMarketing Data Science